IMPLEMENTASI WAVELET TRANSFORM DAN NEURAL NETWORK UNTUK MENGENALI MOTIF BATIK

Oky Antoro, Hartatik
UNIVERSITAS AMIKOM YOGYAKARTA.2017

A B S T R A C T

Batik is one of Indonesian cultures. Batik motif consists of two big categories, i.e. geometric and non geometric categories. Various batik motif which exists becomes challenge in research to develop model which can classify those motif.

This research combines feature extraction methods based on color moments which consist of mean, standard deviation and skewness and spectral decomposition by using Discrete Wavelet Transform (DWT). Whereas to do classification Artificial Neural Network (ANN) classification method is used as classification method. Batik category which will be classified are buketan, ceplok, kawung, parang and truntum.

According to experiment result which compares between usage of haar and daubechies2 wavelet by using 10-fold cross validation obtained highest average accurary after five times experimentation for every configuration is haar wavelet with learning rate 0,2 and momentum 0,2 which produces accuracy 71,2%. Result of the lowest average accurary i.e. 61,2% which uses daubechies2 wavelet with learning rate 0,8 and momentum 0,8.

Keywords - Batik, Color Moments, Discrete Wavelet Transform, Artificial Neural Network, 10-fold cross validation

CategoryUndergraduate Thesis
Posted Date( undocumented )
Modified Date05 Agustus 2017
Download File Publikasi_14.11.7646.pdf
Google Scholar